Microsoft co-founder Bill Gates has warned that artificial intelligence could bring approximately 20 years of economic and social upheaval before its potential benefits are fully realised. Speaking on The Verge's Decoder podcast on 22 October 2026, Gates predicted that AI could increasingly replace office work, while future humanoid robots might take over physical tasks such as construction, agriculture and even surgery. He expects substantial disruption to employment and argues that governments must take responsibility for managing the transition rather than leaving it entirely to technology companies. However, Gates also raised a fundamental question: Even if machines become capable of performing nearly every human task, should society allow them to? He proposed that certain activities, including childcare, cooking, sports and art, could be designated Human Reserved: areas in which society deliberately chooses to preserve human participation, even if machines become technically capable of doing the work. The concept was first outlined in his August 2026 essay on AI risks and governance. Gates also suggested that some communities might consciously limit their use of advanced technology, comparing this possibility to the Amish. The conversation also addressed the economic future of music. Host Nilay Patel pointed out that digital technology has made recordings almost infinitely reproducible, undermining traditional sources of income for musicians. He asked how an economy could function when scarcity disappears, Gates argued that conventional economic incentives might become less necessary in a future of AI-driven abundance. However, he offered no specific solution for sustaining musicians' livelihoods during the potentially turbulent transition. This leaves an important question unanswered: even if society eventually decides that music-making should remain a human activity, how will professional musicians survive until then? Gates's proposed protections remain ideas for public debate, not government policy. His 20-year timeline is an estimate, not a scientifically established forecast, and there is no certainty that AI will acquire all the capabilities he anticipates.